1 BANGKOK TRAFFIC MONITORING SYSTEM Sorawit NARUPITI Associate Professor Transportation Research Laboratory, Department of Civil Engineering, Chulalongkorn University Phayathai Road, Bangkok Thailand Fax: Takehiko KATO Sumitomo Electric Industries, Ltd 1-1-3, Shimaya, Konohana-ku, Osaka, Japan TEL: FAX: Hajime SAKAKIBARA Sumitomo Electric Industries, Ltd , Sekiguchi, Bunkyo-ku, Tokyo, , Japan TEL: FAX: Napon SRISAKDA Graduate Student Department of Civil Engineering, Chulalongkorn University Phayathai Road, Bangkok Thailand Fax: Abstract: Traffic data is essential for determining better traffic management and providing valuable information to drivers. Especially to tackle traffic congestion, traffic data is needed to understand traffic patterns, and to come up with proper solutions for congestion alleviation. In Bangkok, many authorities are involved with the management of traffic, and in recent years, Royal Thai Police has installed a comprehensive image processing detector system throughout Bangkok, in order to monitor traffic. This paper describes Bangkok traffic monitoring system of Royal Thai Police. The system consists of 152 image processing cameras installed at 45 intersections and central unit at a traffic control center. The detectors are capable of providing traffic volume, speed, queue length, and traffic image. All the cameras are linked to the Central Traffic Control and Command center, where the data are stored and made available to operators and commanders. A set of traffic data from the system was analyzed to display the traffic patterns in Bangkok. The system could provide the knowledge of traffic both in details (lane-by-lane, by intersection) or by aggregate (general traffic patterns). The analyses demonstrate that the information from Bangkok traffic monitoring system in useful for both planning and operation. Some ideas on future work are also presented. Key Words: Traffic Monitoring, Data Collection, Bangkok 1. INTRODUCTION Nowadays, several organizations focus on the mitigation of traffic congestion especially in Bangkok. The knowledge of traffic is required first to understand, and to come up with appropriate and pertinent solutions for improving traffic congestion. Traffic data are obtained from various kinds of traffic sensors, traditionally installed on roads. In recent years, the technology of image processing have been adopted in the field of traffic data collection in which the video camera can be installed on the road section or intersection to collect traffic data such as traffic volume, speed, and occupancy. The collected traffic data are generally used to analyze traffic condition and further provide traffic information in order to manage traffic. Among many traffic data collection technologies, video image processing is one of the
2 promising and mature technologies that could be used for capturing traffic flow condition. The ability of video image processing in collecting traffic data has been demonstrated and accepted in practice worldwide. Not only it is important to obtain many traffic data through traffic data collection, but also it is of importance to convert the received traffic data into useful traffic information comprehended and suitable for traffic condition interpretation. The basic traffic data can be analyzed to yield useful interpretation. Basic traffic characteristic can show the amount of traffic demand over time, and the level of demand on various locations. Traffic performance measures, such as queue length, can be a valuable indicator to show the traffic operational conditions. Moreover, the related information such as traffic signal status and operation is also a valuable parameter that could be linked to the performance of the traffic control, and illustrates the potential improvement to traffic control operations and traffic situation. Royal Thai Police has installed the traffic detection system in Bangkok, and traffic data have been collected and processed at Traffic Command and Control center. Traffic data are collected and processed using image processing cameras at 152 locations throughout the road network in Bangkok, at 45 intersections. In the same project, 77 CCTVs have been installed for traffic surveillance purpose. The project can be seen as a large scale implementation of the advanced traffic data collection technology. In addition, the command and control center has been updated to integrate many parts of the center s functions together, such as the integration of Phase 1 and 2 CCTV to the new central surveillance system and traffic report from traffic data processing (image processing) at 152 detection locations in Bangkok. This paper reports the development of traffic monitoring system in Bangkok and the analysis of collected data from the system. First, the brief description of traffic monitoring system is explained. Then the principle of traffic data collection is presented. Details on data collected from the system is displayed. Next, a sample of traffic data are analyzed to determine the patterns of traffic in Bangkok. The data are also analyzed to exhibit the performance of traffic flow and traffic control. The paper demonstrates the use of traffic monitoring system to monitor traffic conditions, and the capability to draw results on traffic flow and control conditions in Bangkok. 2. TRAFFIC MONITORING SYSTEM CONFIGURATION AND FUNCTIONS 2.1 System Configuration As shown in Figure 1, the system is configured with the following six components: 1) Image Processing Type Detector The camera is installed on roadside pole with the height of 8-10 m above road level, as an example shown in Figure 2. The locations of the camera are near stop line looking at the approach to the intersection. One camera covers a maximum of 4 lanes for traffic volume collection and a maximum of 2 lanes of queue length measurements. Traffic data that could be collected from the camera are traffic volume, vehicle type, speed, and image (still picture) data.
3 Collection Server Storage Server Web Server LAN Communication Control Server Web Client PC Image Processing Detectors Figure 1. Traffic monitoring system configuration (a) Lacation (b) Detection area of a camera Figure 2. Image Processing Type detector and its installation Traffic data are sent back to the center via ADSL lease line. The system is designed to optimize the communication rate and thus the 256 kbps line is sufficient for the linkage between each roadside camera and the center. The traffic data (volume, speed, queue length) transmission is set to every 5 minutes, while the image data is sent every 1 minute and this transmission interval can be changed at the center. 2) Communication Control Server This communicates with image processing detector terminals using TCP/IP to request, retrieve, edit and control a variety of data and send data to the Collection Server. 3) Collection Server This holds and controls parameters (constants) required to operate the system and supervises the operation of the Communication Control Server and converts the format of data received from the Communication Control Server to transfer the data to the Storage Server.
4 4) Storage Server This stores and controls data received from the Collection Server in a specified format and distributes stored data upon a request from the Web server. 5) Webserver All the functions of the operation terminal are realized by application on the web server. Then, the operation is possible by access from the web browser on the operation terminal. The Web server gets data from the storage server by request of the operation terminal, and generates web page for web browser. 6) Web Client PC The operation terminal is a Web based system designed to process traffic data in Bangkok by importing image data from image processing cameras. This data can be analyzed and displayed on the map and graphs shown in the Figure 3. Figure 3. Example of display from the monitoring system (Web Client PC)
5 2.2 Functions The Bangkok traffic monitoring system aims at enhancing the traffic management in Bangkok. It provides knowledge of traffic to the traffic operators and commanders. The traffic information can be used for planning and operational aspects. For planning analysis, the traffic data can be used for reviewing the historical traffic conditions. Traffic volumes are reviewed to get the actual traffic demand patterns which would lead to examination of the past traffic management strategies, finding traffic hot spots (critical junction/area), and the periods of traffic congestion. This could help the operator (police) to plan the workforce for their manual operation. At the same time, the traffic information can be used for detailed traffic analysis to determine the systematic traffic management solution, such as the optimization of traffic signal control, lane management, or the implementation of traffic management measures. For operational analysis, the traffic monitoring system provides timely and continuous traffic surveillance. The monitoring of the changes in traffic conditions is crucial for real-time traffic management decision. At this time, city of Bangkok does not possess urban traffic signal control (UTC) which could automatically control traffic signals in the road network in realtime. Nonetheless, the traffic monitoring system gives valuable traffic conditions for operators and commanders at the center. The continuous surveillance enables the operators to explore the change in traffic conditions, the change in traffic hot spots, and the effect of congestion of one intersection to adjacent nodes. The operator can not only judge the right execution through numerical data, but also quickly scan the current and historical traffic conditions through a series of traffic images. In the case of the enhancement of traffic signal control system from the traditional one into the adaptive traffic signal control system (i.e. UTC), the data from the traffic monitoring system is also readily used for such a system. Since the image processing detectors are situated at the stop lines, it is an ideal location for gathering traffic data input for the advanced traffic control system. 3. TRAFFIC DATA ANALYSIS The traffic monitoring system enables grasping the 24 hours traffic conditions on in Bangkok city. The continuous surveillance makes available useful traffic information. In this section, the current traffic conditions analyzed using traffic data from the system are explained. The traffic monitoring system collects important traffic data, namely traffic volume, vehicle types, speed, and queue length. With these data, analyses can be done to disclose useful historical and current traffic conditions (real-time). 3.1 Traffic Distribution Traffic volume is collected on every traffic movement at an intersection continuously. Therefore, the trend in traffic volume can be summarized. The analysis can reveal temporal distribution of volume over time, by each intersection or for all intersections. Figure 4 shows the aggregate traffic volume entering all intersections (in which the system is installed). The figure shows the peak traffic demand on the road network in Bangkok. Based on the data on Friday July 2, 2011, the total amount of entering volume at all installed
6 Figure 4. Hourly distribution of traffic in Bangkok (All intersections) intersections is the highest at 7:00-8:00am. Bangkok roads keep busy throughout the day, mainly from 6:30-19:00. The hourly distribution can be uncovered for each intersection as well, to show the difference in traffic patterns among spatial locations. Figure 5 shows a sample of hourly distribution at each intersection. The historical traffic demand is essential for estimating specific traffic pattern and for planning and preparation of the man-power for manual traffic management. It is also the valuable database for performing traffic prediction, when the real-time current traffic condition and historical traffic database are used as input for short-term traffic forecast. Figure 5. Hourly distribution at each intersection (aggregate entering volumes) The traffic volume can be aggregated for weekly distribution as well. Figure 6 shows an example of weekly distribution of traffic. The figure is used for pinpoint the busy day(s) at the intersection and it discloses different traffic patterns by days of week.
7 Wednesday June 30, 2010 Thursday July 1, 2010 Friday July 2, 2010 Saturday July 3, 2010 Sunday July 4, 2010 Monday July 5, 2010 Tuseday July 6, 2010 Figure 6. Hourly distribution by day of week (Example, Asoke intersection) From Figure 6, at this intersection, it is observed that the morning peak periods on Monday, Tuesday, Wednesday, and Thursday starts at 6:00am and sharply rise. This implies a lot of traffic is on the roads during this peak period. The peak traffic on these days reach their peak before 7:00am as compared to other days (Friday, Saturday, Sunday) when the peak periods start at 10:00am. In the evening, the traffic drops during 18:00-19:00pm. There is not discernable difference between days of week. This is intuitive plausible as this intersection is at busy business and night life area. This analysis can be done for many weeks and the historical trend can show that each day of week has a specific hourly distribution. Figure 7 exemplifies the display on traffic volume on each approach at one intersection. The picture (or the numeric data) showing the demand by approach is useful to learn the congestion pattern at the specific intersection. It can be used for priority treatment at each intersection. From the figure, at XX intersection traffic on approach I-DET 68 (in-bound) is heavy in the morning and traffic on approach I-DET 66 (out-bound) is heavy in the afternoon. The data analysis can disclose the critical movement of traffic at the intersection which varies by time of day. 3.2 Determination of peak periods For grasping traffic flow pattern at peak periods, the time of highest traffic volume at intersection are pointed on the maps. The traffic flows into the city center in the morning and the peak of traffic volume are occurring earlier at entrance intersections than those in the city center. The traffic flows out from the city center in the evening and the peak of traffic volume are occurring earlier at exit intersections than those in the city center. (This tendency is more distinctive than the morning). This traffic flow pattern means that the managing of border intersections between city center and suburb is important. Figure 8 summarizes the peak patterns of traffic at all intersections. The picture gives a valuable view on the movement of traffic demand (and congestion) on the network.
8 Figure 7. Example of traffic volume at an intersection Figure 8. Peak periods at each intersection
9 3.3 Detailed traffic analysis on lane-by-lane basis The advantage of Bangkok traffic monitoring system is the collection of data on the lane basis. This enables us to examine the traffic demand and use of each lane of the approach. Figure 9 exemplifies a graph showing the data at lane-by-lane level. Figure 9. Example of lane-by-lane data (Sukhumvit 71 on 06/07/2010) Generally, the lane-by-lane analysis is not performed when only the overall picture of traffic condition is needed. Nonetheless, the detailed analysis on lane usage reveals the lane utilization. From Figure 9, graph lines with different colors indicate traffic volume on each lane. With the knowledge on the turn restriction on each lane, the graph lines can be used to determine whether the use of lane is optimal. 3.4 Analysis on balanced and unbalanced lane usage To quickly understand the characteristics of lane usage, the comparison of traffic volumes among lanes in the approach can be considered. A balance of lane utilization can be observed. The balanced is defined as traffic volumes on two more lanes in the same direction are similar or not. If so, a balanced is called, otherwise unbalanced. Figure 10 illustrates the method for identifying the balance usage. From data analysis at 45 intersections, traffic characteristics fall into unbalanced lane usage during peak periods in most intersections. This suggests that the signal operations as well as the lane regulation must be considered in order to get better traffic flow efficiency.
10 3.5 Estimated Cycle Length Figure 10. Balanced vs. unbalanced lane usage This data can be used for estimating the cycle time of a signal at each intersection. Since this traffic data collection system is independent with the traffic signal control system, the period of green time and red time of each signal phase and a cycle length of the signal at the intersection is not stored or reported. Nonetheless, the cycle time can be estimated from the traffic movement at the beginning of two green time periods. An example of the result of cycle time estimation is shown in Figure 11. Figure 11. Example of Cycle length estimation From the database, it can be seen that 20 intersections (Ex. Kiakguy Int., Sang Hee Int., Sapan Dang Int. and Sawankalok Int.) are manually controlled with around 500 seconds of cycle length and most of other intersections are controlled with around 400 seconds of cycle length in the peak periods.
11 3.6 Discussion on Bangkok traffic The analyses of Bangkok traffic data reveal the characteristics of Bangkok traffic. Moreover, the data collection is carried out continuously and the data can be retrieved in real-time. The data show that each intersection has its own traffic patterns. The traffic database can give accurate traffic parameters for a specific location. Generally, the traffic volumes at most intersections are heavy, leading to congestion for a long period of time of day. Queue monitoring shows that queues are accumulated on road sections, implying that the road network experiences oversaturation during those time periods. Since the system is not connected to the traffic signal control system, the signal timings (cycle time and green time) are not normally recorded. Nonetheless, the cycle length can be estimated based on queue formation characteristics. The signal timings are crucial for determining roadway capacity (service flow) and the collection of signal timings can show the performance of signal operation. As found from the traffic analyses, the duration of cycle time at 45 intersections in Bangkok is long. In common practice in Western countries, the maximum duration of the cycle time is seconds. In Thailand, however, the duration of cycle time of 400 seconds or more is commonly seen. 4. IDEAS FOR FUTURE WORK The installation and use of Bangkok traffic monitoring system gives rich data collection and valuable information for traffic operation. The system can give many pieces of traffic analyses and knowledge on network-wide road traffic condition. Moreover, the system estimates and records the actual operations, such as the duration of cycle time from the manual operations by police officers. One observation from the data analysis is that the manual control by police results in very long cycle length. The field observation reveals that the police tend to give mercy extension of green time to traffic which is currently receiving green duration (Figure 12). As generally known, too long cycle time causes excessive delay and long queue to entire intersection operation. The traffic monitoring system can detect long cycle time and may be used for coordinating the manual operation not to have too long cycle time (preferably 300 seconds or less). Figure 12. Common mercy extension of green in manual traffic signal control
12 The traffic monitoring system is capable of obtaining knowledge of traffic throughout the network. This is beneficial in that a strategic plan for network-wide traffic control can be carried out. With the ability to accurately predict traffic movement and congestion patterns on the network, the network-wide traffic management can efficiently be conducted. - Amount of traffic can be predicted based on the level of traffic in the surrounded area and historical traffic patterns - Traffic control can be coordinated so that it prepares capacity for the predicted traffic. For instance, for morning peak period, the amount of traffic at the city center can be predicted knowing the amount of in-bound traffic at border intersections. Operations at intersections would aim to increase capacity to heavy traffic directions. Likewise, for the evening peak period, the border intersections can prepare capacity to accommodate amount of traffic outbound from the city center. As the locations of image processing detectors are at the stop lines of intersections and the capability of collecting traffic data in real-time, the data from detectors could be used as input to any adaptive traffic signal control system. In Bangkok, a demonstration project was conducted in September 2009 at Kiek Kai intersection. The adaptive signal controller was installed at this intersection and the system utilized the data from the detectors. The project demonstrated that the traffic monitoring system could also be further utilized as data collection for adaptive signal control. 5. CONCLUSION This paper describes the traffic monitoring system which has been implemented in Bangkok by Royal Thai Police. The system provides traffic surveillance at 45 intersections covering a large area of road network in Bangkok. The 145 image processing detectors collect traffic volume, speed, vehicle type, queue length and traffic image and the data are gathered at a central control center. The system is capable of collecting historical data as well as real-time monitoring by traffic operators. The system provides data as well as tools to disclose knowledge of traffic. A sample of traffic data from the system was obtained and analyzed. Amount and distribution of traffic flow can be obtained. From the Bangkok traffic data, traffic on streets are heavy throughout the day. The system helps identify the locations of congestion and traffic patterns. The historical data give understanding on the trends and effects of traffic management. The analyses can be done at lane-by-lane or aggregate intersection basis. Some important notices about Bangkok traffic can be found from Bangkok traffic analyses. Traffic behaviors such as lane utilization (balanced or not) is observed. The cycle time of signal by manual control is generally long, and this can cause inefficient traffic operations. This gives some ideas for further development. The real-time data collection can be used for determining proper traffic management strategies, which consider network-wide traffic movement and prediction. The signal control can be improved through adaptive signal control utilizing the data from the monitoring system.
13 ACKNOWLEDGEMENTS The authors would like to acknowledge the Royal Thai Police (Traffic Police Division and their staffs) in supporting this activity. They have best collaboration and allow us to access data and access to Traffic Control and Command Center. REFERENCES Faculty of Engineering, Chulalongkorn University (2011). Bangkok Traffic Analysis using Police Image Processing Camera Data. Final Report. Faculty of Engineering, Chulalongkorn University (2010). Report on Adaptive Signal Control Demonstration Project. Submitted to Royal Thai Police. Siwarak Unsiwilai (2010). Effectiveness of Adaptive Signal Control to Traffic Characteristic Study Case : Kiak Kai Intersection. Unpublished Civil Engineering Project, Department of Civil Engineering, Chulalongkorn University. Sumitomo Electric Industry. Traffic Flow Monitoring System User s Manual.
ASSET MANAGEMENT FOR KANSAS COUNTIES: THE STATE OF PRACTICE by KEVIN D. FRIEDRICHS B.S., Kansas State University, 2006 A THESIS Submitted in partial fulfillment of the requirements for the degree MASTER
PROJECT FINAL REPORT Grant Agreement number: 212117 Project acronym: FUTUREFARM Project title: FUTUREFARM-Integration of Farm Management Information Systems to support real-time management decisions and
Transportation Research Part C 15 (2007) 380 391 www.elsevier.com/locate/trc Evaluation of a cellular phone-based system for measurements of traffic speeds and travel times: A case study from Israel Hillel
WHITWORTH, JEFFREY N., M.S. Applying Hybrid Cloud Systems to Solve Challenges Posed by the Big Data Problem (2013) Directed by Dr. Shanmugathasan (Shan) Suthaharan. 66 pp. The problem of Big Data poses
7 th Framework Programme Theme 7: TRANSPORT (including AERONAUTICS) CONDUITS, Coordination Of Network Descriptors for Urban Intelligent Transport Systems Contract n 218636 I. Kaparias, M.G.H. Bell email@example.com
Incident Management Index Purpose Description Relevance for Large Scale Events Options Technologies Impacts Integration potential Implementation Best Cases and Examples 1 of 13 Purpose An incident is an
` City of Beverly Hills Public Works Services Department 345Foothill Road Beverly Hills, CA. 90210 Infor (Hansen) Computerized Maintenance Management System Request for Proposal For Infor System Re-implementation
The Industrial Internet@Work Marco Annunziata & Peter C. Evans Table of Contents Executive Summary The Industrial Internet Towards No Unplanned Downtime 3 Introduction A New Information and Collaboration
Business Travel eprocurement Business Travel Process analog or web based? A study by PricewaterhouseCoopers Unternehmensberatung GmbH Frankfurt/Berlin, November 2000 This report has been created by PricewaterhouseCoopers
FRAUNHOFER RESEARCH INSTITUTION AISEC CLOUD COMPUTING SECURITY PROTECTION GOALS.TAXONOMY.MARKET REVIEW. DR. WERNER STREITBERGER, ANGELIKA RUPPEL 02/2010 Parkring 4 D-85748 Garching b. München Tel.: +49
I. Automatic Vehicle Location and the Potential for Pupil Transportation Applications Automatic vehicle location (AVL) is an automated vehicle tracking system made possible by navigational technologies
Problem Management Contents Introduction Overview Goal of Problem Management Components of Problem Management Challenges to Effective Problem Management Difference between Problem and Incident Management
Big Data in Cloud Environments Satoshi Tsuchiya Yoshinori Sakamoto Yuichi Tsuchimoto Vivian Lee In recent years, accompanied by lower prices of information and communications technology (ICT) equipment
Measuring Efficacy of Information Security Policies A Case Study of UAE based company Muhammad Sohail Qureshi Department of Computer and Systems Sciences Degree project 30 HE credits Degree subject (Information
Series 2 ORGANIZATIONAL MANAGEMENT Financial Management STRUCTURE Management Information Systems SUSTAINABLE DEVELOPMENT Impact Assessment Career Development Strategic Planning SYSTEMS Supervision Objectives
TIME-VARIANT TRAVEL TIME DISTRIBUTIONS AND RELIABILITY METRICS AND THEIR UTILITY IN RELIABILITY ASSESSMENTS Patricio Alvarez, Universidad del Bío Bío, firstname.lastname@example.org Mohammed Hadi, Florida International
Management of patient information Trends and challenges in Member States Based on the findings of the second global survey on ehealth Global Observatory for ehealth series - Volume 6 WHO Library Cataloguing-in-Publication
Industrial Internet: Putting the vision into practice Industrial Internet business models for machine and component manufacturers What Industrial Internet & Industry 4.0 mean The benefits of connecting
CYBERSECURITY WORKFORCE DEVELOPMENT MATRIX RESOURCE GUIDE October 2011 CIO.GOV Workforce Development Matrix Resource Guide 1 Table of Contents Introduction & Purpose... 2 The Workforce Development Matrix
Monitoring and Diagnosing Applications with 4.0 Mark W. Johnson IBM Corporation The (Application Response Measurement) standard provides a way to manage business transactions. By embedding simple calls
IBM i on Power - Performance FAQ February 5, 2013 IBM Corporation Table of Contents 1 Introduction 8 1.1 Purpose of this document 8 1.2 Overview 8 1.3 Document Responsibilities 8 2 What Is Performance?
Ville-Pekka Peltonen Software Asset Management Current state and use cases Helsinki Metropolia University of Applied Sciences Master of Business Administration Master's Degree Programme in Business Informatics
Prepared for Virtual Hold Technology June 2006 The Total Economic Impact Of Virtual Hold s Virtual Queuing Solutions Project Director: Dan Merriman TABLE OF CONTENTS Executive Summary...4 Purpose...4 Methodology...4
CRM: Taking One-to-One Marketing to the Next Level An Executive White Paper Coravue, Inc. 7742 Redlands St., #3041 Los Angeles, CA 90293 USA (310) 305-1525 www.coravue.com Table of Contents Introduction...1
Wireless Enterprise Innovations: Raising Issues Of Temporality And Activities Per Andersson, Mats Edenius, Christopher Rosenqvist, and Pablo Valiente Stockholm School of Economics, Center for Information
NETWORK MONITORING AND DATA ANALYSIS IN WIRELESS NETWORKS By Yongjie Cai A Dissertation Proposal Submitted to the Graduate Faculty in Computer Science in Partial Fulfillment of the Requirements for the
HP Performance Engineering Best Practices Series for Performance Engineers and Managers Performance Monitoring Best Practices Document Release Date: 201 Software Release Date: 2014 Legal Notices Warranty